Anthracyclines are always included in the treatment regimen for diffuse large cell lymphoma, but the role of anthracyclines in treating patients with symptomatic FL is not clear. Both R-CHOP and R-CVP are used frequently in the management of symptomatic patients with FL. Hiddemann et al.  illustrated current advances in treatment modalities for follicular lymphoma including the use of chemotherapy, radio-therapy, monoclonal antibodies, vaccine strategies, and stem cell transplants. Liu et al.  had highlighted increased survival time over the past 25 years for those who have stage 4 FL with advances in treatment options. It is still not clear whether addition of anthracyclines in the therapy for FL confers any benefit.
In our analyses we tried to answer the question whether addition of anthracyclines to the chemotherapeutic regimen improves response in patients with FL. To compare the response rates between R-CHOP and R-CVP in a prospective fashion would require a multi-year multicenter trial involving a large number of patients. For example, using nQuery 5.0, one can determine the sample size per treatment group if the hypothesized treatment difference is that obtained in the current study between R-CHOP and R-CVP for CR+PR (96% vs. 81%). Assuming 80% power using 2-tailed Fisher's Exact test, with significance at p < 0.05, the trial would require about 75 patients per treatment group, not counting replacement of patients with incomplete data. With this in mind, we conducted the current meta-analysis of all relevant literature comparing the two treatment options, the primary outcome measures being CR, PR, or CR+PR. Our sources were the commonly used search engines and the abstracts of the American Society of Hematology. In the first analysis, comparing R-CHOP and R-CVP as frontline agents, only two pertinent articles could be retrieved, whereas four were available for the second analysis. This paucity of studies underlines the difficulty inherent in systemically comparing response rates in patients with indolent diseases treated with two different but very effective regimens.
As shown in table 4, patients treated with frontline R-CVP had a much higher chance of developing a CR compared to those with frontline R-CHOP. However, when combined with studies of relapsed or previously treated patients, the difference–still favoring R-CVP–became only marginally significant. One possible explanation for this result is that the patients treated with R-CHOP had higher ECOG scores, were older, and were women. On the contrary, in both the analyses, the probability of achieving an overall response (CR or PR) was significantly higher when patients were treated with R-CHOP as opposed to R-CVP. With the data presented, it would seem that patients would fare better overall with R-CHOP, for 96% experienced some form of response in both analyses, significantly higher than the 81% found for patients treated with R-CVP. However it is unclear and we were unable to obtain usable data to test whether a response of any type necessarily correlates with increased survival.
Multiple observations need to be made about the present data. The most obvious is that the definitions for CR and PR are not uniform across studies. Second, we could retrieve only one prospective study involving R-CVP  in the treatment of FL. Third, the high CR rates reported by Czuczman et al.  and Domingo-Domenech et al.  were not supported by Hiddemann et al. . In the study by Czuczman et al.  the number of patients treated with R-CHOP was 38 and in Domingo-Domenech et al.  was 16, approximately 20% of the total patient population treated with R-CHOP in our analyses. Hence, the CR rate from this R-CHOP group was clearly lower than R-CVP. From the combined data it would seem evident from both studies that R-CVP was overall inferior to R-CHOP. This difference could be from selection bias and non-comparability of the study subjects.
No data were located for comparing R-CHOP vs. R-CVP for relapsed patients with follicular lymphoma, and no data for previously treated and relapsed patients on R-CVP. Finally, it was unfortunate that for the second analysis only age, sex, and stage were reported as demographic and clinical variables in all four studies. We had to ignore the values for patients' performance status, bone marrow involvement, ECOG status, LDH status, B-symptoms, and IPI because none of these variables was available for all four studies in the analysis. Editors may wish to require future studies to report such fundamental variables not only to satisfy readers' immediate interests but as well with an eye towards future meta-analyses.
For indolent disease like follicular lymphoma, response may not be an adequate end-point. Although some recent studies suggest the importance of CR for survival, the progression-free survival or time to treatment failure (TTF) could be more relevant and could avoid the bias of measurement. Marcus et al  investigated the addition of rituximab to CVP (R-CVP) therapy compared with CVP therapy alone and showed a significant advantage for R-CVP for remission rate (81% vs. 57%; P < .001), TTF (27 months vs. 7 months; P < .001), and time to next therapy (median not reached vs. 12 months; P < .001). However, remission rates and TTF achieved by R-CVP appear comparable to the results obtained by CHOP alone. A substantially better outcome seems to be achieved by R-CHOP. Adverse effects, particularly severe granulocytopenia, were less frequently encountered after CVP (14%) or R-CVP (24%) than after CHOP (53%) or R-CHOP (63%).
An unavoidable weakness of any meta-analysis is its inability to perform multivariate analyses that might throw light on the importance of various potentially confounding variables for the overall outcome, in ways not avail to any of the constituent studies because of their limited sample sizes.
QUOROM provides a system for rating studies to be included in meta-analyses of randomized clinical trials. We did not use this system, for we wished to incorporate all available data. We judged all four cited articles as equally relevant in providing accurate data for our purposes. Other studies utilizing G-CSF with R-CHOP for the treatment of FL, such as by Niitsu et al., were excluded because they attempted to use G-CSF as a treatment modality, and responses changed depending on the dose of G-CSF provided. As with any paper, much information was omitted from our study and one should not use this article as a crutch when determining the appropriate chemotherapeutic protocol for a patient. Profiles of side effects of Adriamycin were not evaluated, nor could we provide a correlation between specific responses and length of survival or cost of treatment. One should always evaluate specific cases when deciding the treatment protocol appropriate for the individual.
The international PRIMA study testing the efficacy of maintenance therapy by rituximab may provide important data in the field of the best induction in patients with follicular lymphoma.
In summary, we conclude that treatment for patients with FL should be individualized. R-CHOP and R-CVP protocols can both achieve excellent overall response. In patients with known cardiac history, omission of anthracyclines is reasonable, and R-CVP provides a very good CR rate. In younger patients with FL where cumulative cardio-toxicity may be of importance in the long term and in whom future stem cell transplantation is an option, R-CVP may be a more appealing option. How the response rates translate to survival is not known and certainly needs to be further clarified in prospectively designed long-term follow-up studies.